HALL. — THERMAL AND ELECTRICAL EFFECTS IN SOFT IRON. 625 



magnesium, " light calcined magnesia," on the other hand, was found 

 to bear the heating test well, losing at first a small amount of weight, in 

 drying, probably, but remaining constant in weight at high temperature 

 therealter. 



Accordingly we undertook to repack our Thomson effect apparatus, 

 using now the dried magnesium oxide, and to repeat the experiments 

 on lateral flow through the packing as well as those on the temperature 

 coeflicient of the thermal conductivity of iron. 



As we had a strong apprehension that our boilers would now leak vapor 

 of water or of naphthalin into the packing space, though we had done 

 what we could to stop such leaks as we had discovered, and as we feared 

 that any condensation of vapor in the powdery oxide would be highly 

 objectionable, we put a sort of buffer of asbestos fibre about 0.5 cm. 

 thick against the boiler at each end of the space to be packed, trust- 

 ing that the upward movement of air through this layer would carry 

 off vapor coming into it. Then, after putting in place the wrapping of 

 magnesia and asbestos which lies outside the outer guard-ring (see 

 Figure 8), we poured the magnesium oxide down into the space between 

 the many bars, striving, by tapping the outer covering and by pressing 

 down the oxide with strips of metal, to get the light powder so well 

 packed that it would not by subsequent settling leave empty spaces 

 under the bars. The density of the powder as thus placed was about 

 0.17 gram per cubic centimeter. 



After this we made, in spite of a considerable amount of leakage of 

 naphthalin, especially toward the last, observations which resulted as 

 follows : 



Lateral outflow from each main bar per 1° difference of temperatu)-e 

 between the main bars and the inner guard-7-ing, 



0.00791 cal. with both pots at 100°, July 17-18 

 0.00894 " " " " " " " 27 



Temperature coefficient of thermal conductivity in iron between 115° 

 anc? 204°, with reference to the value at 115°, 



f = - 0.00076, July 19 and 24. 



It is to be noted that in this packing the thermal conductivity 

 appears to be about 30 per cent greater at 218° than at lOO"". 



VOL. XLII. 40 



